Controlling Polymerization of β-Amyloid and Prion-derived Peptides with Synthetic Small Molecule Ligands*

The Alzheimer β-amyloid peptide (Aβ) and a fragment of the prion protein have the capacity of forming amyloid-like fibrils when incubated under physiological conditions in vitro. Here we show that a small amyloid ligand, RO-47-1816/001, enhances this process severalfold by binding to amyloid molecules and apparently promote formation of the peptide-to-peptide bonds that join the monomers of the amyloid fibrils. This effect could be antagonized by other ligands, including analogues of RO-47-1816/001, as well as the structurally unrelated ligand Congo red. Analogues of RO-47-1816/001 with low affinity for amyloid did not display any antagonistic effect. In conclusion, these data suggest that synthetic molecules, and possibly also small natural substances present in the brain, may act in a chaperone-like fashion, promoting Aβ polymerization and growth of amyloid fibrils in vitro and possibly also in vivo. Furthermore, we demonstrate that small organic molecules can be used to inhibit the action of amyloid-enhancing compounds.

The amyloid deposits in brain vasculature and parenchyma that are the main histopathological hallmarks of Alzheimer's disease (1) are composed of large polymers of Alzheimer ␤-amyloid peptide (A␤). 1 This peptide is present in two major forms, one being 40 amino acids long, and the other, more aggregable form being 42 amino acids long (2)(3)(4). The A␤ peptide is secreted by numerous cell types in the body but the amyloid deposits are only present in the central nervous system.
The amyloid deposits are formed by the aggregation of individual monomers of A␤ peptide into very large polymers which have a fibril-like appearance when observed in an electron microscope (5). These polymers can also be observed by light microscopy following staining with certain histological dyes such as Congo red and thioflavine T (6).
Soluble A␤ peptide can be detected in blood and cerebrospinal fluid. The levels are, however, very low, usually in the low nano-to picomolar range and at these concentrations the pep-tide polymerizes at a very slow rate (7). A number of recent studies on the mechanisms of amyloid formation have conclusively shown that the presence of preformed oligo-or polymers of the amyloid peptide in the reaction mixture increases the polymerization rate dramatically (7). These multimers serve as templates for the reaction and, as a result, the initial, slow phase of primary nucleation is eliminated.
It has also been proposed that charged molecules, such as gangliosides (8) and metal ions, including Zn 2ϩ and Cu 2ϩ (9), may enhance formation of amyloid in vitro and in vivo. Apolipoprotein E4, a well established risk factor for Alzheimer's disease (10) has also been suggested to enhance amyloid formation by serving as a "pathological chaperone" (11)(12)(13). In the case of apolipoprotein E, there are conflicting data suggesting that the protein can enhance as well as inhibit amyloidogenesis (14,15).
The existence of putative co-factors capable of enhancing amyloid formation potentially offers new targets for pharmaceuticals for use in the treatment of amyloid disorders. However, so far very little is known about how such compounds might work, and it is therefore difficult to develop assays for finding candidate pharmaceuticals and to predict their molecular mode of action.
A number of molecules which function as amyloid ligands have been described. These include the textile dyes Congo red and thioflavine T that bind to amyloid fibrils but not to nonpolymerized peptide. These compounds, therefore, have found widespread use in histopathological diagnosis of amyloid disorders. The elements in the amyloid fibril that these compounds bind to are probably the novel surfaces or binding pockets generated as a result of amyloid peptide polymerization (16). In addition to these small organic molecules there are also peptide-based ligands. The rationale behind such ligands is that during polymerization short stretches of amino acid residues interact and form ␤-strands that join the monomers encompassing the fibril (17)(18)(19). It was speculated that short peptides corresponding to one of the strands in the ␤-sheet can bind to the amyloid peptide/protein. With this approach, ligands capable of interfering with amyloid formation in vitro and possibly also in vivo have been identified (20 -23).
Here, we first investigated if nonpeptidic, synthetic molecules of low molecular weight can have similar effects as the larger natural substances described above. By having access to molecules that easily can be altered structurally, the mechanism through which small molecules regulate the higher order structure of the A␤ peptide can be probed with a more rational, structure based, approach than otherwise. Moreover, small synthetic molecules with pharmacokinetic and toxicological properties allowing them to be easily administered to animals, may also be useful in studies of amyloidogenesis in vivo.
Analysis of A␤ Peptide Polymerization-This assay was carried out essentially as described previously (24). Briefly, 96-well plates (Maxisorp, Nunc) were coated with peptide by incubating them with a solution of A␤ 1-42 or A␤ 1-40 (2.5 M) in Tris-buffered saline (50 mM Tris/ HCl, pH 7.4, 150 mM NaCl, and 0.05% NaN 3 ). To each well, 100 l of the solution was added, and the plates were incubated at 37°C with shaking for 48 h. The peptide solution was then decanted. Staining with a solution of Congo red showed that the polymeric peptide had bound to the wells. After removal of the nonbound peptide, the plates were allowed to dry. Coated plates were stored at Ϫ20°C in a desiccator until used. On the day of experiment, the plates were blocked by addition of 300 l of PBS containing 0.05% (v/v) Tween 20 (PBS-T) and 1% bovine serum albumin/well for 2 h at room temperature. The plates were then washed with PBS-T, and the fluid was decanted. Biotin-A␤ 1-40 or biotin-A␤ 1-42 was dissolved in Me 2 SO and diluted in Tris-buffered saline with NaN 3 (0.05%) to a final concentration of 20 nM. The plates were incubated 3 h, unless stated otherwise, at 37°C with shaking. Nonbound peptide was removed by washing the plates three times with PBS-T (300 l/well). Streptavidin-peroxidase was diluted with PBS-T containing 1% bovine serum albumin and added to the plates (200 l/well). After incubation (30 min at room temperature) the solution was flicked off, and the plates were washed four times with PBS-T. Tetramethylbenzidine was used as chromogenic substrate for the peroxidase. After termination of the reaction with sulfuric acid (0.33 M), absorbance was measured at 455 nm using a SpectraMAX 250, 96-well plate reader. Nonspecific binding is defined as binding of biotin-A␤ to wells that had not been coated with A␤. There was a linear relationship between peroxidase activity and the amount of peptide bound (data not shown). Nonspecific binding was, on average, approximately 15% of total binding (data not shown).
Fluorescence Equilibrium Binding-In these experiments, the test compounds (0.5 M) were incubated in white 96-well plates in the presence of polymerized A␤   that this compound bound to A␤ 1-42 fibrils (Fig. 1B) with a K d of approximately 6.2 M (Fig. 1C). In the same type of experiments it could also be shown that the compound binds to A␤ 1-40 with a similar affinity (data not shown).
RO-47-1816/001 Enhances Binding of Soluble A␤ to Immobilized Polymers-Here, an assay was employed where binding of biotinylated A␤ 1-40 or A␤ 1-42 to immobilized nonlabeled A␤ 1-42 was used as a measure for A␤ polymerization (24). When RO-47-1816/001 was added to the assay mixture, polymerization increased severalfold ( Fig. 2A). As seen in the figure, the compound also increased binding to control wells to which no nonlabeled peptide had been attached, but only to a small extent. A reasonable explanation is that the test compound increased binding of soluble biotinylated A␤ 1-40 to the small amounts of biotinylated A␤ 1-40 that had been nonspecifically absorbed to the plastic material of the wells.
By coating the plates with A␤ 1-40 , instead of A␤ 1-42 polymers, and studying binding of biotin-A␤ 1-40 , virtually identical results were obtained (data not shown). This demonstrates that the two C-terminal residues of A␤ 1-42 are not required for the effect of RO-47-1816/001. The results also show that the compound requires immobilized polymers for its effect in the assay and that the increase in biotin signal reflects enhanced incorporation of soluble peptide into the immobilized polymers. We also studied how addition of RO-47-1816/001 affected the resistance of the polymers to trypsin (26) and to solubilization with high concentrations of urea. Addition of 30 M RO-47-1816/001 did not have any significant effect on these parameters, suggesting that polymers formed in the presence of the compound had a similar or identical sensitivity toward the protease and chaotrope solubilization. Because the amyloid formed in the presence of RO-47-1816/001 shared these features of genuine amyloid (i.e. relative resistance to proteolysis and chaotropes), it is likely that it is very similar, or identical, to amyloid formed in the absence of the compound.
It has previously been demonstrated that at least a fraction of polymeric A␤ remains associated when separated under the relative harsh denaturing conditions employed in SDS-PAGE (27). We used this feature of the peptide to study if addition of RO-47-1816/001 to nonlabeled A␤ 1-40 increased formation of high molecular weight forms of the peptide. In the absence of RO-47-1816/001, immunoreactive bands with apparent molecular masses of approximately 4, 8 and 16 kDa, probably corresponding to A␤ mono-, di-, and tetramers, respectively, were obtained (Fig. 2B). Addition of RO-47-1816/001 increased formation of the 16-kDa form dose-dependently and induced formation of even larger polymers. This directly demonstrates that the compound enhances association of A␤ into large multimers.
In a similar experiment, nonlabeled A␤ was incubated with increasing concentrations of RO-47-1816/001. After 19 h of incubation at 37°C, it was observed that the compound dosedependently increased the amount of polymerized peptide as measured by Congo red staining (28), suggesting that the compound had stimulated formation of fibrils fulfilling this criteria for amyloid.
Finally, we also studied if addition of RO-47-1816/001 affected the morphology of fibrils formed by nonlabeled A␤ 1-40 and A␤  . The peptides were incubated in the presence of 30 M RO-47-1816/001 for 20 h and then subjected to electron microscopy as described previously (24). As seen in Fig. 3, addition of the compound to the polymerization mixture had no apparent effect on the morphology of the fibrils (Fig. 3, compare  A with B and C with D). A reasonable interpretation of this experiment is that RO-47-1816/001 enhances polymerization without affecting the ultrastructural appearance of the A␤ fibrils.
RO-47-1816/001 Increases the Rate of A␤ Polymerization-The binding of biotin-A␤ to immobilized polymers is time- dependent (24,29). We therefore decided to study how RO-47-1816/001 affects polymerization at various time points over an interval of 5 h. As seen in Fig. 4, incorporation of biotin-A␤ 1-40 was essentially linear over time in the studied time interval in the absence of RO-47-1816/001 in agreement with previous studies (30). Addition of the compound changed this pattern, most notably in the presence of the highest concentration used (50 M) where maximal binding was observed already after 1-h incubation. A reasonable interpretation of these data is that the compound increases the rate through which soluble peptide is incorporated into the immobilized polymers.
RO-47-1816/001 Enhances Binding of Soluble Prion Protein Fragment to Immobilized Polymers-Here, the specificity of the compound for A␤-amyloid was addressed. A fragment of the prion protein corresponding to amino acids 106 -126 (PrP 106 -126) spontaneously polymerizes into amyloid-like fibrils, morphologically indistinguishable from those formed by A␤ (31)(32)(33). First, it was studied if RO-47-1816/001 can bind to the prion fragment. As seen in Table I, RO-47-1816/001 bound to PrP 106 -126 with an affinity similar to that displayed for A␤  , demonstrating that RO-47-1816/001 is not a specific A␤ ligand but also binds other types of amyloid. Using the same type of assay as employed for A␤, the effect of RO-47-1816/001 on PrP 106 -126 polymerization was investigated. In Fig. 5 is shown that the compound dose-dependently increased biotin-PrP 106 -126 binding. Its potency was, however, lower here than in the A␤ assay (compare Figs. 2A and 5). The conclusion is therefore that the effect of the compound is not specific for A␤-amyloid and thus can be used to enhance the polymerization of at least one other, structurally unrelated, amyloid peptide.
A␤ Ligands Structurally Related to RO-47-1816/001 Can Antagonize Its Amyloid Formation-enhancing Effect-A selection of five analogues of RO-47-1816/001 (Fig. 1A) were tested for their affinity toward polymerized A␤   (Table I). Three of the compounds bound to A␤ with an affinity similar to that of RO-47-1816/001. The two others had low affinity for the peptide (RO-47-3537/000 and RO-65-5780/000) with approximately 10 times higher K d . Clearly, the two compounds with the lowest affinity are less lipophilic than the four with the highest affinity. There are also other structural differences. The polymerization accelerating compound RO-47-1817/001 is an ester, all the other molecules are amides. In addition, the phenyl ring of RO-47-1816/001 is unsubstituted, whereas the antagonists carry methoxy or benzyloxy substituents.
When tested in the A␤ polymerization assay, none of the RO-47-1816/001 variants had any clear effect on biotin-A␤ 1-40 binding (Fig. 6, left panel). This was surprising considering that: (i) three out of five compounds bound to the amyloid with an affinity in the same range as RO-47-1816/001 and (ii) they had striking structural similarities to RO-47-1816/001. This prompted us to investigate if the analogues, by competing with RO-47-1816/001 binding, interfered with its capability to enhance polymerization. As seen in Fig. 6 (right panel), three of the five tested compounds dose-dependently reduced binding of    (Table I).
Congo Red Inhibits the Effect of RO-47-1816/001-Congo red is an amyloid ligand that changes its spectral properties upon binding to amyloid fibrils and therefore is a useful tool in histochemical diagnosis of Alzheimer's disease and other amyloid disorders (6). It has previously been shown to interact with the central core region of A␤ as well as to regions with similar structural properties in other amyloid-forming proteins and peptides (16). Structurally, Congo red is very different from the hydrophobic pyridone derivatives described here (Fig. 1A). Fig. 7 shows the effect of Congo red on the binding of biotin-A␤ 1-40 both in the presence and in the absence of RO-47-1816/ 001. In the absence, Congo red slightly enhanced binding at the highest concentrations employed, which is in agreement with previous published data (29). In the presence of 10 M RO-47-1816/001, Congo red reduced binding similar to that of the previously analyzed RO-47-1816/001 analogues. As seen in the figure, Congo red only reduced the effect of RO-47-1816/001 and was not capable of reducing polymerization below that obtained in the absence of RO-47-1816/001. At the highest concentrations used, the level of binding was essentially identical to that in the absence of RO-47-1816/001. This suggests that RO-47-1816/001 and Congo red interact with A␤ at the same binding site. DISCUSSION Here it has been shown that RO-47-1816/001, a synthetic compound with a molecular mass of 592 Da accelerates polymerization of amyloid peptides several times its own size. There are only a few known examples of small molecule compounds regulating the state of polymerization of proteins. The most well known compounds of this class are probably colchicine and paclitaxel (previously taxol), that induce disassembly and assembly, respectively, of microtubuli and are used clinically to treat gout and neoplasms (34).
Binding studies showed that RO-47-1816/001 displayed affinity both to A␤-amyloid as well as to amyloid formed by a peptide derived from the prion protein (Prp 106 -126 ). This is a feature shared by several other organic compounds, including Congo red and thioflavine T. It is believed that the novel surfaces generated as a result of the adoption of ␤-pleated sheet structures during polymerization serve as binding sites for these amyloid ligands (35). This property of the compounds is widely used to histochemically distinguish amyloid from normal tissue. Because Congo red could interfere with the action of RO-47-1816/001, it is reasonable to assume that the latter interacted with the same binding site(s) as the former and that this site(s) therefore can be of importance for putative amyloidenhancing compounds and their antagonists.
It was interesting to observe that analogues of the amyloid ligand RO-47-1816/001 failed to produce an accelerating effect FIG. 8. Proposed mechanism to explain how the amyloid ligand RO-47-1816/001 accelerates incorporation of soluble A␤ peptide into preexisting polymers. A, soluble A␤ directly interacts and binds to the immobilized polymer (indicated as a trimer in the figure). In addition to the sites were the peptide-to-peptide interactions take place, there is a binding site for the ligand RO-47-1816/001. After the ligand has bound to the A␤ polymer, either a novel surface is formed (a composite of the ligand and the polymer), to which the soluble peptide can bind or alternatively, the protruding part of RO-47-1816/001 per se binds the soluble peptide and cross-links it to the polymer. It is not clear if the ligand remains associated with the polymer (lower part of the figure) after the soluble peptide has bound to the polymer or if it dissociates from the peptide complex (upper part of the figure). B, the antagonists also bind to the polymers but fail to cross-link the polymer to the soluble peptide. Instead it will occupy the binding site and inhibit the binding and the effect of RO-47-1816/001 but without affecting the direct peptide-to-peptide interaction. on amyloid formation despite having similar affinities to A␤ as the original compound. This initially surprising finding was later explained when it was revealed that the inactive ligands instead antagonized the effect of RO-47-1816/001. The present finding therefore suggests that RO-47-1816/001 exerts its action not only by binding to the amyloid polymer but that it also has an additional effect that the antagonists do not possess. It is tempting to speculate that RO-47-1816/001, after it has bound to the amyloid fibril, binds to the soluble peptide and brings it into close proximity to the fibril (Fig. 8), thereby facilitating docking of the peptides via their specific binding sequences (20,21,36). Whether cross-linking is mediated directly by RO-47-1816/001 or by a composite surface of RO-47-1816/001 and the A␤ polymer is not clear. The latter mechanism is, however, more favored in nature. The most well known example is probably the interaction of the T-cell receptor with its antigen. The affinity of the receptor for a peptide antigen per se is low but when bound to and presented by the major histocompatibility complex, the affinity increases dramatically. Under these circumstances, the T-cell receptor can bind both to the antigen and the major histocompatibility complex (37,38), and the increased number of contacts leads to increased affinity.
There are also examples of natural compounds being used as pharmaceuticals that employ this strategy in mediating their specific action. Cyclosporin A and FK506 both inhibit calcineurin by forming a complex with cyclophilin and FK506binding protein, respectively, that display high affinity toward the phosphatase (39,40). Recently, this mechanism was elegantly used to increase the affinity of a bifunctional ligand designed to induce heterodimerization of the FK506-binding protein and an SH2 domain (41).
It is not clear if the ability to enhance polymerization of amyloid peptides is a common feature in small organic molecules or if it is restricted to a very small number of substances. We do, however, want to point to the possibility that clinically used pharmaceuticals may have this property and, at least theoretically, may promote amyloidogenesis in vivo.
It can be speculated that the present results suggest that the brain might contain small endogenous compounds having similar effects as RO-47-1816/001. Apolipoprotein E (apoE) and gangliosides have been mentioned previously, but it is still possible that there are other natural compounds with similar effects. An interesting observation by Selkoe and co-workers (42) is that Congo red stabilizes monomeric A␤ in cell culture supernatant, which may suggest that compounds with similar effects as those of RO-47-1816/001 may be secreted by cells. Finally, the finding presented here, that some compounds, including Congo red, displaying affinity for amyloid can antagonize the action of amyloid enhancers, may therefore be of therapeutic significance.